#GWGWorking on a bunch of WordPress plugins for Indieweb stuff. Trying to simultaneously write for myself and a little for others
#jgarberGWG: Cool! Let's pick this conversation up again (particularly regarding a DC IWC). Hope you (and everyone else here) have a pleasant evening (or afternoon or morning depending on where you are)!
KartikPrabhu, chrissaad, davemenninger, mdik and gRegor` joined the channel
#aaronpkhuh, bridgy seems to have not found my latest post.
#mkoben_thatmustbeme: I've been working on an indie-auth'd accessible "Dashboard" view as my default browser homepage, but I've been having problems with IndieAuth losing the authentication. >.<
#mkoI'm pretty sure it's a config issue in how I'm serving cookies right now, but it's blocking me from it being actually useful.
#ben_thatmustbememko, you could just set your own cookie after auth, then its all just your management of it
#aaronpkyeah indieauth has no concept of "sessions" or being logged in, it's just to identify the user
#mkoYeah. I do. I think it's a bug in the Node module that I'm using for auth cookies. I'm going to rewrite it to work more consistently.
#mkoOh, I'm not blaming IndieAuth at all. I actually meant to say "with losing the IndieAuth authentication" but apparently transcribed it when I was typing. I'm a bit scattered this morning.
#ben_thatmustbemesyncing files to a place to show the css styling thus far, which is pretty much just all ripped out of firefoxOS
#tantekaaronpk - I proposed precisely that two days ago in the W3C Social Web WG - that "x is following y" should be a post, just as "x favorited/liked y" is a post. :)
#LoqiTo follow is the concept of establishing a digital relationship to another person or entity so that you can receive updates from them over a given social media channel http://indiewebcamp.com/follow
#aaronpk(I am also simultaneously regretting segmenting my URLs by post type)
#tantekrestrains himself from specific comments on URL design back in the day when we last discussed this ;)
#aaronpkI should have just done date URLs for all posts in a single namespace, rather than articles, notes, replies, metrics, etc
#ben_thatmustbemeI'm thinking of just making my archive controller redirect to a specific post though so they can be treated either way
#tantekaaronpk - I believe you made a URL design decision based on how you thought your readers would want to primarily consume your content (by type).
#aaronpki'd have to actually rename all my files on disk in order to change my URLs
#aaronpkbut now that I have a database it's much easier to have multiple feeds by post type or other things like I do with tags
#aaronpkben_thatmustbeme: that's what I would do right now except I might have notes/2014/10/10/1 and replies/2014/10/10/1 and they are different
#tantekben_thatmust: I have /YYYY/DDD/TN where YYYY-DDD is ISO ordinal date, T is single character post type, and N is the incremental number of that post type for that date.
#tantekand you can trim characters from the end of the URL and it does the right thing
#aaronpkeither way on mine I want to make the levels represent directories
#aaronpkI do know I want to show my "main" feed on my home page, not all posts
#aaronpkand I also know I want to have different feeds for certain types, like a "notes" feed separate from a "checkin" feed
#aaronpkbut I feel like that should be a queryable thing, not baked into the way I store my files
#ben_thatmustbemeheh, tantek, I love MVC for that, model doesn't necessarily mean i have to store in a DB, just a translation from somewhere to my code... i'm thinking I just make my models pull/write to on disk files, but then cache in DB
#aaronpkI also generate a lot of posts per day, so I do want the day in the URL too
#tantekaaronpk - I do post-type-specific feeds by query (parameters in URLs)
#tantekI made the choice of date-primary over type-primary because dates are a known thing with a known range, whereas types are still being figured out. Thought that was a no brainer in terms of designing which should be more primary in URL / storage design (that's supposted to be longterm, thus more stable).
#aaronpkoh another consideration of my URLs is that I often create metrics posts out of order (some data is imported hours after it's created, when I may have created several other posts in the mean time)
#aaronpkthat's the reason I use HHMMSS instead of numeric index for those posts
#tantekaaronpk - that seems like a sensible approach
#kylewmtantek: I meant the components included in the url more than the order+exact layout
#kylewmwhen I said that I based them on yours and aaron's
#tantekkylewm: huh - I think the order is important
#aaronpkand if there's a conflict then I just artificially increment the "seconds" part by one until there's no conflict
#kylewmtantek: yeah order is definitely important, and I like your rationale for date first ... had to come to that conclusion on my own
Pierre-O joined the channel
#aaronpkso I think I'd want to support that format still too. I would end up with two URL formats: YYYY/MM/DD/N and YYYY/MM/DD/HHMMSS (and also YYYY/MM/DD/N-slug)
#aaronpkonly potential danger there is if I somehow had over a million regular posts in a single day. but that seems unlikely. 999,999 ought to be good enough for anybody.
benjamin-agaric joined the channel
#aaronpkoops no. over 100,000, so I am limited to 99,999 per day
snarfed joined the channel
#tantekis amazed at aaronpk's *personal* big data problem.
#tantekkylewm: how do we document these URL design learnings in a way that suggests a pattern for future implementers?
#aaronpkI guess I should write a post like kylewm's based on all my irc chatter here
#aaronpkoh man here's another great example of why you should have date-based permalinks with full path navagability
#aaronpkhttp://indiewebcamp.com/permalinks has a link to my tweet from 2009 about URL shorteners. I wanted to find the version of that post on my site, but in 2009 I didn't have a citation on my tweets
#Mark87You have multiple applications that weren't designed to interoperate and store their data in different ways and formats, and you wrap them in adapters and make them talk to each other, or present a higher level adapter that aggregates their functionality to make "one" application
#Mark87_I ask because the realm of indie websites could be thought of in some ways as similar, and I just wonder if theres anything to be learned from that connection
#Mark87For now we have webmentions and microformats to connect us
#Mark87And I guess they serve as our wrappers/adapters
#aaronpkright, webmentions and microformats connect websites which otherwise have completely different backend and incompatible implementations
#LoqiCORS is an acronym for "cross-origin resource sharing," a mechanism for allowing browsers to make JavaScript requests to fetch resources from other domains http://indiewebcamp.com/CORS
#KartikPrabhuwell hopefully webmentions and mf2 are improved not thrown away. SOmething like how HTML is never thrown away
#KartikPrabhudoes not like the Javascript hammer gfor everything
#aaronpkI'm assuming you were implying there may be new things added to the list in the future, like the recent work on actions
#tilgovithe biggest thing I want to change is that it's a bit of a black box right now
brianloveswords joined the channel
#tilgovigood for the blogger who doesn't want to write anything but add annotations to their site. and good for the individual who wants the bookmarklet/extension in their browser for annotating other sites.
#KartikPrabhuyour source has so much stuff before the content...
#tilgovinot so good for the indie web developer who wants a script with an API like webmention.io style
#cweiskeah, now I know why I didn't install it. python
#tilgoviKartikPrabhu: yeah, progressive enhancement. It's 99% a single page client app, but there's a small static template there for the webmention
#KartikPrabhutilgovi: well if you are prog. enhancing. maybe only load on request or something... anyway not crucial
#tilgovicweiske: if we had the server API well documented and made it easy for you to build the front end without any Python knowledge, would that help? Then you could just implement the REST endpoints for storage.
#KartikPrabhuDoing good. lot of actual work stuff rather than indieweb
mlncn joined the channel
#tilgovicweiske: yeah. right now that use case is covered somewhat well, but only if you want to take it as a "black box" and let us run the service or if you want to run the whole stack yourself.
#tilgoviAnd I'm quite sure I may have messed things up, anyway, since this is my first time trying this out.
#tilgoviIt seems to work on aaronpk's site, but not sure about others'p
#tilgoviI think it worked on yours too when I tried it last night
#tilgovibut it just showed up as a mention at the bottom and the summary wasn't good
#tilgoviI may have improved it by the time I slept, but didn't try it again
#tilgoviobviously, though, the fragmention would be the best :-D
#KartikPrabhutilgovi: yeah it showed up as a mention becasue it found a URL but no microformats. so it couldn't decide if it was a reply or not
#tilgoviI'll give it a shot right now. I definitely had it very broken at first.
#tilgoviAlso I'm so sorry for any nonsense I'm leaving on anyone's notes as I test broken things. I looked around but didn't find whether anyone's built a validator I can use to test.
#KartikPrabhusure. feel free to add a "u-in-reply-to" and also a fragmention (if you want) and resend the mention
#KartikPrabhutilgovi: interesting. I seem to have picked up some microformats from that page already but didn't seem to pick up a "in-reply-to" URL even though it exists... I'll debug this evening
#tilgoviis it going to be a problem if there's more than one in-reply-to?
#tilgoviI'm tempted to have one that's just the URL for the header/metadata part of my posts and one that's on the quote itself that has the fragmention
#tilgovii'll remove the non-fragmention one for now
#KartikPrabhutilgovi: hmm good point! I don't know :P You can update the post as you want and I'll update my code to treat it nicely :)
#tantekaaronpk, re: another great example of why you should have date-based permalinks with full path navagability, see also the write-up of archive-based dated-link repairing from last Thursday! http://indiewebcamp.com/irc/2014-10-23#t1414105046134
Mark87 and bengar joined the channel
#doriantantek: absolute numbers in pagination schemes while yer at it
#tantekdorian - as in don't do them - they don't produce re-referenceable results
#tantekalso a topic I remember debating with aaronpk 2-3 years ago in IRC
#aaronpkben_thatmustbeme: I'm talking about two separate schemes. the first, for articles/notes/photos/checkins/etc is just an incrementing index starting at 1. I'm limited to 99,999 that way. The other, for metrics, is always HHMMSS and I can have up to 86400 of those.
#aaronpkI guess i'll have to have a special handler for how to deal with 23:59:59, probably just start searching backwards for a free slot
#joskartantek: Wouldn't the "fetching-permalink-from-archive" thing work for most use-cases even though the date is not embedded in the permalink? (since I guess you already have timestamps for when you linked to it in your posts / embedded replies / whatnot)
#tantekjoskar - "you already have timestamps for when you linked to it in your posts" - that assumes information *outside* the URL itself. That's my point. purely by the *URL* without any side-information you can attempt recovery etc.
#tantekless work, less overhead for such recoverability
#joskarI can see why it is more fragile to trust your own timestamp than the permalink itself. I personally wouldn't worry about "less work" since domain takeovers very seldom happens thought
#Loqijoskar meant to say: I can see why it is more fragile to trust your own timestamp than the permalink itself. I personally wouldn't worry about "less work" since domain takeovers very seldom happens though
#danlykejoskar, I haven't been following this thread as carefully as I perhaps should have, but when you say "domain takeovers very seldom happen" do you mean "we rarely lose our own domains" (probably true), or are you talking about the domains we link to (got many examples of that).
#joskardanlyke: I meant "very rarely" more as "not frequent enough to care about speed of execution". I acknowledge that domain deaths happens all the time, but probably won't affect you more than a couple of times per year.
#joskartantek: I guess my point of view is "It is easier to change your own habits than to change others." I don't think that forcing a permalink-format onto people is going to work. (although you can encourage the use of one)
#joskar(that said, I still think it's an interesting concept)
#tantekjoskar: you're right about changing your own habits being easier. hence what I was suggesting was a practice for indieweb site owners to do themselves with their own URLs first, and second, local-link-wrap any URLs they link to that don't follow the year/date pattern.
#KartikPrabhutilgovi: yes. I meant without hypothesis storing stuff for me